I’ve noticed in every platform of “training” software I’ve ever used that I continually “break” the algorithm for “power duration”.
I thought I’m good at math. I’m an engineer for a living and use it often. I didn’t think it was mathematically possible to have an average over a longer duration be HIGHER than an average over a shorter duration.
Example from today:
Time, power
1, 355
2, 354
5, 254,
6, 244
10, 250
12, 231
20, 234
So, in that example…I broke their algorithm twice in the same workout. The 6 minute number should be higher or equal to the 10 minute and the 12 minute number equal or higher than the 20min.
Like, WTF.
Explain please. I’m curious. It’s this way with TP, my Wahoo, a Powertap app thingy. If you export the data file to an Excel doc, you can do a formula to drag down and the formula gets it correct.
This messes me up because I pay attention to trends of these over time.
I’ve noticed in every platform of “training” software I’ve ever used that I continually “break” the algorithm for “power duration”.
I thought I’m good at math. I’m an engineer for a living and use it often. I didn’t think it was mathematically possible to have an average over a longer duration be HIGHER than an average over a shorter duration.
Example from today:
Time, power
1, 355
2, 354
5, 254,
6, 244
10, 250
12, 231
20, 234
So, in that example…I broke their algorithm twice in the same workout. The 6 minute number should be higher or equal to the 10 minute and the 12 minute number equal or higher than the 20min.
Like, WTF.
Explain please. I’m curious. It’s this way with TP, my Wahoo, a Powertap app thingy. If you export the data file to an Excel doc, you can do a formula to drag down and the formula gets it correct. This messes me up because I pay attention to trends of these over time.
I think those numbers are your avg power through each X min. From 6-10 min, you increased your power output such that your avg for the first 10 min went up from 244 for 1st 6 min to 250 for 1st 10 min. Then you went down a bit from 10-12 min, then back up a bit from 12-20 for the 20 min avg of 234 watts. That’s the way I would interpret your data, rather than your “I broke their algorithm…”.
You’re selecting the max average for Xmins from a variable signal.
For say an interval session (but the same for any variable data set), it’s possible for a shorter duration to include a larger proportion of easy to hard effort than a longer one, meaning it will have a lower average. So while you certainly COULD physically work harder for a shorter period, the max power for the shorter period will be lower because in reality, you didn’t.
A worked example is best to see this. Let’s do a hypothetical case where your results for 6 and 10mins would occur:
Say you’re doing 1min intervals at 399W with 5min recoveries @213W
Your best 10 minute average will include two 1 min efforts, and 8 minutes of recovery
(2399+8213)/10 = 250.2W
Your best 10 minute average will only include 1 min effort, but still needs another 5 minutes which have to come from recovery periods
(1399+5213)/6 = 244W
No matter where you pick your 6 mins you can’t include another 1min effort. So 6 mins is the worst of all worlds. Anything below that will include a higher proportion of interval to recovery. Anything greater will include additional interval time. So selected from this continuous 1min on 5min recovery session your results would be as follows:
Obviously, unless you’re using an ERG, you’re not going to get such consistent figures and if it’s a race or unstructured cycle it’ll be much harder to calculate, but the rationale is the same.
This is what these guys are trying to explain in pictorial form
If you’re doing very “on-off” efforts such as trainer intervals then your MMP for the interval + rest will (correctly) be a localised minimum in your Power-Duration curve and it will (seemingly counter-intuitively) rise again at longer durations as you start to include effort from the next interval.
I was thinking about it as a non-time-bound dataset incorrectly. Given it moves over time, that makes much more sense.
You can’t just ts say that “I averaged 250 for 20min so must have also for 15minâ€. If you remove 5min you may remove the power contribution to that value.
